Ap0c08E1IT RESONE

ID 166 221 TN 008 058

TITLE Manual for the. Bateria de Examines de Aptitud General'.(BERG) . Section II. Development.

INSTITUTION' Employment of Training Administration (DOL),Washington, D.C. Office of Research,andDevelopment.

PUB DATE 77NOTE . 30p.AVAILABLE FROM Superintendent of Documents, United States Government

Printing Office, Washington, D.C. 20402 (Stodk Number029-014-000103-1)

EDRS PRICE 0MF-$0.83 HC-$2.06 Plus Postage.DESCRIPTORS Adults; *Aptitude Tests; *Equated icores;. *Norms;

Occupational Tests; Sampling; Senior High Schools;*Spanish; Spanish Speaking; Test.Coistruction;Testing;iTest'Interpretatiou; TeSt /tems; *VocationalAptitude

IDENTIFIERS *Bateria de Examines de Aptitud General; *GeneralAptitude'Test Battery; Test Manuals

os.

ABSTRACTThe development and norming of a Spanish language

edition of the United States. Employment Service (USES) GeneralAptitude Test Battery (GATB) is described. The newedition, calledthe Bateria de Examenes de Aptitud General (BEAG), was designed toreplace an earlier tt4nslation, the BGPA, prepared for use in PuertoRico. Forms A and B were developed to have identical item contentwith the GATB except for Part 1 (Name Comparison) rand Part 4(Vocabulary) which required Spanish, names and vocabulary rather thanEnglish. Experimental procedures, norms samples" and results aredescribed for an administration of-the tests tc bilingual groups ofhigh-school seniors and adults with less than high school education.Item analysis ,indicated a number .of items in Part 4 wereunsatisfactory. A revised edition was prepared and a secondcomparability study was, performed. Another study was performed tocompare the BEAG with_the BGPA. (CTM)

***********************************************************************'* Reproductions supplied by EDRS are the best that can be made

from the original document.*********************************************************************#*

Haneseral [BEG]

:ion

14.*

i, f..A

At

U S DEPORTMENT OF HEALTH..EQUCATION L WELFARE4ATIONAL INSTITUTEOF

EDUCATION

H4frDOCUMENT HAS BEEN REPRO-ED EXACTLY AS RECEIVED FROM

RSON OR ORGANIZATION ORIGIN -IT POINTS OF VIEW OR OPINIONS

1:11DDO NOT NECESSARILY REPRE-0 4ICIAL NATIONAL INSTITUTE OF

ON POSITION OR POLICY

o0 '1 '00,

a 0.o

'0 ap -0o -0 coo 0

,.,0 0 0 0:

N

--ManuaBateria de Examenesde Aptitud General [BEAG]

SeCtionDevelopment

U.S. Department of LaborRay Marshall, Secretary

Employment and Training AdministrationErnest G. Green

, Assistant Secretary of LaborforEmployMent and Training1977

3

O

ACKNOWLEDGMEN SDevelopment 'of the Bateria de Examenes e Aptitud General

(BEAG) was done in cooperation with the follow' g,State employmentservices:' Arizona, California, Colorado, District o Columbia, Florida,Idaho, Illinois, Louisiana, Michigan, Minnesota, Nev da, New Jersey, New

, Mexico, Ndw York, Oregon, Puerto Rico, Texas, ashington, Wisconsin,and Wyoming.

Because of its continuing interest in test de elopment for Spanish-speaking applicants, the California agency partic pated in planning andconducting all phases of the test development e ort to develop a newSpanish language edition of the GATB, includin preparation of the ex-perimental designs and coordidatiNn of the data ollection. Much of thecredit for the successful completion of the proje is due to the efforts ofLillian Avery, Manager, and Margaret Walker, J hn Gonzalez, Raid Na-zario, and Margaret Scott of the California Test esearch Field Cente.

tiThe New York agency, in addition to the r participation in various

phases of the BEAG development, summarized the results of the nation-wide tryout and made recommendations for the final edition. This was .adif6Cult task competently done by Albert Vega ank Dominick Carminatiof the New York agency.

A key role in the developmentlof the EAG was played by personnel of the Puerto Rico agency. Their pvtici ation in several phses of /the research made it possible to make the B G acceptable' for use .byPherto Rico and mainland testing personnel w o_had previously used theBGPA. Particular mention is .made of the fine ontributions of Gladys V.,Colon and Zenaida Rivera, who helped in devel pingohe final consensus inareas where it was difficult to.hchieve agreement in'the working of the di-rections for administration.

Special thanks are due Patricia Ramos hnd Socorro Castafieda ofthe California Test Research Field Center who spent untold number ofhotrits typing, editing, and proofreading the BEAG manual and testmaterials.

AGRADECIMIE TOLa Bateria de Examenes de Aptitud Gen ral (BEAG) fue desa-

rrollada en cooperaci6n con los siguientes servicio de empleo estatal: Ari-zona, California, Colorado, District of Columbia, lorida, Idaho, Illinois,Louisiana, Michigan, Minnesota, Nevada, New Je ey, New Mexico, NewYork, Oregon, Puerto Rico, Texas, Washington, Wisconsin y Wyoming.'-

Debido al interes constante en el desarrollo de examenes para lossolicitafnes de hatila hispana, la agencia de California particip6 en elplaneamiento y direccion de todas las fases del esfuerzo del desarrollo deexamenes para la elaboracion. de una nueva edicion de la GATB en la

iii

.. .,

lengull espanola. Esto incluyo Ia preparation de disefios de experiment° yla coordinaciAn de _coleccion de datos. Mucho merit° del exito de estepr ecto finalizado se debe a los esfuerzos de Lillian Avery; Gerente, yM et. Walker, John Gonzalez, Raul Naiario y Margaret Scott del"California Test Research field Center".

La agencia de Nueva York' adenias de su participaciOn en variasfases del desarrollo de la BEAU, resumi6 los resultados del experimento

' Nevado a cabe por toda la nation e hizo recomendaciones para Ia edici6nfinal. Esta fue una tarea dificil competentemente concluida por. Albert Vegay Dominick Carminati de la agencia de Nueva York.

El personal de Ia agencia de Puerto Rico desempefio un papel im-, portante en el desarrollo de Ia BEAG. Su participation en varias.fases de la

investigation sistematica y cientifica biz° posible que el use de Ia BEAGfuera aceptable por el personal de examenes en Puerto Rico y los EstadosUrjdos Continental, quienes anteriormente habian usado la BGPA. Men -cion especial se debe a las contribuciones excelentes de Gladys V. Colon yZenaida Rivera, quienes asistieron en Ilegar a un consentimiento final enlas areas donde era dificil lograr un acuerdp relativo a las palabras de las'instrucciones para la administration.

Gracias especiales se les debe a Patricia Ramos y a Socorro Cas-tafieda del' "California Test Research Field ('enter', quienes ttabajaronhoras incalculables escribiendo a maquina, repasando y haciendo correc-ciones de prueba al manual de Ia BEAG y a los materiales de examenes.

5iv

FOREWORD (GATB)The General Aptitude Test Battery (GATB) was developed by the

United States Employment Service, and has been used since 1947 by af-filiated state employment service offices. Since that time the GATB hasbeen included in a continuing program of research to validate the testsagainst success in many different occupations. Because of its extensive re-search base, the GATB has corfte to be recognized as the best validatedmultiple aptitude test battery in existence for use in vocational guidance.

The GATB has proved to be a valuable tool in the measurement ofaptitudes of job applicants and of candidates for occupational training.However, many Spanish-speaking individuals who need vocational coun-seling and remedial services to help them become employable and whoare monolingual Spanish-speaking, or.wffb have limited knowledge of the'NEnglish language are not able to take the GATB because of a languagebarrier.

In 1955, research was conducted by Puerto Rico to develop aSpanish language edition of the GATB for use with their population. Thisresearch resulted in'the Bateria General de Pruebas de Aptitud (BGPA).In 1965, the California State Employment Service initiated new researchto develop a Spanish language version of the GATB for use with the Span-

ish-speaking population in the Southwestern United States. The Californiaproject was later expanded to developing a translation suitable for use withthe .Spanish-speaking population throughout the United Slates mainlandand Puerto Rico. Comparability and tryout research studies, involving theparticipation of twenty state employment service agencies, including PuertoRico, were carried out under the auspices of the United States EmploymentService during the period 1968 to.1975.

This Manual presents the Spanish language edition of the GATB,the Bateria de Examenes de Aptitud General (BEAG).

The Bateria de Examenes de Aptitud General (BEAG) Manualispublished in two separate sections, as follows:

Section I, Administration and Scoring, contains the procedures foradministration and scoring of the BEAG, and conversion of theraw test scores to aptitude scores.

Section II, Development, contains technical information on the de-velopment ,of the BEAG.

Users of the BEAG should also have the following sections of the.GATB Manual for information on norms and their development;

Section II, Norms, Occupational Aptitude Pattern Structure, showsthe GATB Occupational Aptitude Pattern Structure which is usedfor counseling purposes.

6

Section III, Development, contains technical 'information on thedevel6Prfient of the GATB.

Many schools and other organizations, both public and private,have been authorized to use the GATB for counseling and research. Infor-mation regarding release of the BEAG for these purposes may be obtainedfrom State Employment Services.

PREFACIO (BERG)La General Aptitude Test Battery (GATB) fue desarrollada por el

Servicio de Empleo de los Estados Unidos, y se ha ushdo desde 1947 porlas oficinas afiliadas del Servicio de Empleo Estatal. tresde ese tiempo, laGATB ha sido incluida en un programa continuo de investigacion sistema-

, tica y cientifica para convalidar los examenes segun el exit° en muchasocupaciones diferentes. A caust de la amplia base de la irivestigaciOn siste-matica y cientifica, la GATB ha sido reconocfda como la mejor convalida-cleft de examenes de multiples aptitudes en existencia para uso corn°. gura

vocacional.

Se ha conwrohado que la GATB es un instrumento importance paramedir las anti trIrs de los solicitantes de empleos y para los cindidatos deinstrticcion o 'clonal. Sin embargo, muchos individuos de habla hispanaquienes necesflah orientacion vocacional y servicios remediales para ayu-darles a ser empleables, y que solamente hablan el espaii61,,o que tienen unconocimiento limitado de la 10igurt inglesa, no pueden tomar la GATB porfalta de comprender el ingles.

En 1955 una investigacion sistematica y cientifica' fue 'llevada acabo por Puerto Rico para desarrollar una edicion de la GATB en la lengua

_espanola para el uso de su poblacion. Esta inve5igaciein result° en laBateria General de Pruebas de Aptitud (BGPM7En 1965 el Servicio deEmpleo Estatal de California inicio unanueva investigacion sistematica ycientifica para el desarrollo de una version de la GATB en la lenguaespafiola para el uso de la poblaciOn de habla hispana en el oeste de losEstados Unidon Mas tarde el proyecto de California fue ampliado paradesarrollar Una traduccion adecuada para el uso de la poblacion de hablahispana en todos los Estados' Unidos y Puerto Rico. Las investigacionessistematicas y eientificas sobre los analisis de comparabilidad y experi-ment°, incluyendo Ia participacion de,veinte fencias de servicio de empleoestatal y Puerto Rico, fueron llevadas a cabo bajo-los servicios del Serviciode Empleo de los Estados Unidos durante los altos 1968 a 1975.

Este Manual presenta la edicion de la GATB en la lengua espariola,la Bateria de Examenes de Aptitud General (BEAG).

El Manual de la Bateria de Ex:tmenes de Aptitud General (BEAG)se publica en dos secciones por separado:

Seccion I, Administracion y CalificaciOn, contiene los procedimien-'tos para administrar y calificar la BEAG, y para Ia conversion de lascalificaciones en crudo del examen a las calificaciones de aptitud.

Seccion II, Desarrollo. contiene inforrnaciOn tecnica sobre el desa-rrollo de la BEAG.

4o

vi

4

, -

Cos que usan la BEAG tamiAin debet litentt las siguientes seecionesdeLManual cicia_GAtlipara informUcion sobre las nornrfat y(hu ciesarrollo.

SecciOn II, Norm'as, la Estillettfra, del Modelo de Aptitud Ocupa--clonal, ensefie Ia Estructuraklel Modelo de Aptitud Ocupacibnal dela GATB, Ia cual seusa Ora los propositos de konsejar.

Seecion III, Desarrollo, contiene informaciOn tecnica so re eLdesa-.rrollo de Ia GATB. /*A muchas escuelask otras organizaciones, tanto publfcas 'cowl

privadas,privadas, se les ha dado liAdtOridad para usar la GATB en aeonsejar in-\Aviduos y para 'lacer investigaciones sobre eke tema ocupacional. Se puede

obtener, del SeviCio de Empleo Estatal, informaciOn en relacion a Ia liber-tad de tiso de Ia BEAG para estostpropOsitos.

6

it?

4,

vii

/

4

'1

4

Y

,

4

CONTENTS

Introduction

i'lranslation rand Tryout of the Bateria de Exam'enes de AptitudGeneral ( BEAG

Broadening Applicability of the BEAG

Comparability of Spanish and English, Language Versions of GATParts-1 and 4 (Study I )

Comparability of Spanish and English Language Versions of GATParisi I and 4 (Study 2 )

Nationwide Operational Tryout of ,the BEAG !-

References

1

Page

3

19

;325

ix

4

i.

INTRODUCTION .

The 1965-66 USES Advisory Panel for OccupationalTest Development. discussed the possible 'need for anew, Spanish language edition of the GATI3 for use inthe 'Southwest. The following oirecommendafion is fromthe Panel's final report:

"On-the basis of information developed withinthe resources of the Panel, i is reco'nlfnendedthat a Spannth edition he derived from thepresent Puerto Rican edition,- with supportingoccupation norms, as a more practical alterna-tive as di ared to a period of instruction inEnglish ''fo owed by use of the standardGAT13.."

In ortr Illbobtain specific and detailed information On

sit

need for a new Spanish language edition of the GAM,a survey was made in June 1967 of fiv Slate Employ-ment' Services in the 'Southwest: California, Arizona,Texas, Colorado, and New Mexico..Resultsfrorn thesurvey indicated dial (I) there were a' great manySpanish-speaking applicants for entry jobs or occu-pinion& training who could not read and understandEnglish well enough to take the GirTB, for mostof these, GATIi testing would be appropriate if a suit-able Spanish ,language edition were available, (3) tilePuerto Rican (edition of the GA1B (BGPA) was notsuitable for most of these applicants, and (4) a newSpanish language edition Of the GATE was neededthe Southwest.

TRANSLATION AND OF THE BATERIA DEEXAMENES DE- APTIUP GENERAL (BE/14)1.

TRANSLATIONIn 1970,44.crsonnel of the California State l'inploymentService prepared a Spanish language. translation of theentire GATI3, including the Manual for the GeneralAptitude Te.vt Battery, 11-1002, Section I: Administra-

. tion and Scoring; Test Booklets 1 and II, Fortitslk and13; Part 8; and the GAII3 Screening Exercises)

TRYOUT PLANA design was developed by the Califorgia agency toevaluate the transfation 'in .the five southwestern Stateswith si7eahle Spanish-speaking populations: Arizona,California, Colorado, Nev Mexico, and. I eXi t,, > EachState was to assume the responsibility of evaluating thetranslated vt_rsn s of the GATI1 Directions for Admin-istration for Parts through I

Fault parcycipating ., tate was to have a sample composi-tion of both nionngual (Spanish-speaking) and bi-lingual, ( those who 14 'poke Spanish and English ) malesand temales.whopassed.the Spanish translation of the

iGA'l'R Screening Exercises, a short screening =test ofrekilijz.ely simple vocabulary and three-dimensionalspice items. Only individuals who had not previously:"been tested with the GATB and who had sullivient edu-cation to understand oral directions and verbal kTntentof the GA 1'13 translation we're to he selected for thestudy.

1

The test administration directions were to he given byexperienced test examiners fluent in Spanish. OnlySpanish-speaking proctors who had training in riATI3test administration were to assist with the study. Dif-

ir ferent methods of conductint., the administration sessionwere suggested by the design in order to encourage theparticipation of the examinees in the' identification ofwords, phrases, and eonceptsunfamiliar to them during

, '. the admigistration of the test:

theIt was suggested that n .-4:rsiamiheo actually perforrwthe exercises to motivatt! them during test sessions.

There was to he no need err scoring the exercises be-cause the research would be concerned~virus' with, the

adequacy of the adAnnistration directions..GATI3 Screening Exercises ( Parts I and 11.).4triwslatedinto Spanish, were to he given to each Mai% iatall;eforeinclusion in the test session. Problems of understand*

were also noted here. ,

lite original design called for each State in the study .toassemble a panel of language experts to evaluate t It c

results of the administration tryouts. In the intrest; , -s;ivlop, time, the des,tyn was modified so that only the.key State (Califo.rnia) wouldassemt?le such a pane'.

i ,

representative of the American Southwest, which could,,evaluate the tryout results from all live Slates and could,',.make final corrections to the translation.

SAMPLE DESCRIPTIONFive States participated in the study, as follows: Cab -[mina (key State), Colorado, New Mexico, Arizona,and Texas. There were a total of 34 sessions: Califor-nia, se0n; Texas, five; New Mexico, six; Colorado,eleven, and Arizona, live, A total of 315 individgals

'were screened by the five States, and 40 of these weredismissed for a variety of reasons. The final sample usedfor the evaluation consisted of 275 individuals, dis-'trihirtethas itutti(,-tited in Table ,

The data in Table I indicate these sample characteris-tics: 1(1) a relatively even representation of males andfemales, bilinguals and monolinguals, (2) a wide rangeof age and (3 ) a low average education, but with sonicrepresentation of higher education levels. It eeins,rea;sonable to conclude that the sample characteristics wereappropriate-for a comprehensive tryout of the new di-.4-ections.

CONDUCT OF THE TRYOUTIn the Spring of 1971, the participating States admin-istered the experimental Spanish language GeneralAptitude' Test Battery to the 34 groups of monolingualand bilingual individuals.A variety pi methods were used to elicit examineeseom-ments and suggestions regarding du; test directions. Insome'sessions examinees were permitted' to.interrupt atwill to ask questions and make comments. In other

--sessions comments were permitted only at the end ofeach exercise o'r onlyrat the end of the testing session,

A specially designed eolunmar.table was developed fortabulation of word, phrase. and concept Variables iden-titiediii the testing sessions, It combined data from the-five States participating in the study. This .table served

Table

Salm* Characteristics and Number of. Cases for Each State(Tryout Study)

State M

NuirtberYlot CasesBain- Mono-

F gual lingual M

Ake

SD Range M

Education'

SD Range

Arizona 20 30 25 25 28.4 101 17-52 7.1' 2.7 3-14'California 49 39 39 49 30.4 8.9 16-50 9, 3.3

Colorado 21 .32 48 5- 33.7 10.0 18-58 8.9' 3.5 . 2-17New Mexico 14 15 29 0 31.3' 12.9 17-63 10.4 4-14Texas `29 26 t27 28. 30.8 11.6_ 17-58 .6.5 2.5 1-12

Total 133 142 I )8 107 30.8 10.6 16-63 8.3. 3.3 1-18

as the raw material for the panel of experts who eval-tiara(' the Spanish instructions for the purpose of modi-fying or cling,* those words and phrases which wereunclear, or misunderstood. It was felt that combiningdata. from the five States on one master thble would

'facilitate the evaluation process that a broad .spec-trum of the entire problem could easily he discerned. Areview of the data singed out no particullir geographical.area which the Spanish language was radically dif-ferent [ruin that in other 'areas. Tlie SouthwesternStates,where Spanish is spoken. read, and written, are(Mite homogeneous in the usage 'of the Spanish lan-guage.

REVIEW PANEL°Assistance from the community in 1.os Angeles was re-quested. A panel of six individuals was chosen Who

'were fluent in Spanish and whidealt, primarily with lieSpanish-speaking population in their daily businessaffairs. It was felt that a compositiOn of such a panelwould offerzan excellent cross section of probleMs en-countefed with individuals of different educationallevels, different environmental factors. and peoples..from countries other than the United States. A furthergoal was that some of the panel rhembers would hifvehad experience' in dealing with people throughoOt thesouthwestern United 'States. As a result of these con-siderations; individuals were chosen from the field of

Radiowilting and commentary in Spanish; News-paperLediting, writing and translating; Unionsnego-tiating and solving problems by direct communicationwith the public segment: Schools- -a teacher with,experience in teaching individuals at all grade levels inSpanish or a combination of Spanish and English: Serv-icestwo persons dealing with the general public inSpanish and' also' with employers and other agencie;.

4

'The educational level of each panel' member was an-other consideration to niiiintain a balance betweenmembers who Were highly proficient in the Spanish lan-guage' and those who, spoke niory nearly like the gen-eral public.'especially the Mexican-American.The panel was assembled for the purpose of 'reviewingthe (lac' collected from. the total sample 'of 275 males...and females of the five-combined States so as to arriveat an acceptable translation. The' panel met kir theinitial *view of the translated instructions into Spanishon July-14, 1 971 aitci completed the overall evaluationof theme instructions on July 29, 1971 in two separatesessions. At the outset, it was agreed among themselvesthat the .evaluation should be' treated in terms of usage

.throughout the.southwestern United States.

CONCLUSIONSA. Screening Procedures

As a screening device, Parts I and 11 proved to'beinadequate. Individuals with a low educationallevel and limited reading ability were screened in.It was usually discovered during the first and sec-ond' practice,exercises of the GATE that such per-,sons were having. great difficulty grasping themeaning of the instructions. Some were ex'usedfrom the test session while others were apparentlykept to avoid embarrassment on their part. The.conclusion that can be draWn from this situationis that, since only one problem of each part musthe correct to pass the screening exercises, the 'actiff guessing correctly any one of the parts couldvery well conic into play, especially if the individ-ual later claimed that he was unable to read theinstructions. The selection based on the screeningexercises was further compounded.by two possible.correet answers to Part 11,' line 2, as reported by

'

12

the'participating states. California recognized thisproblem and resolved it before attempting to screenfora test session.An interesting result of this research study was thefact that, within the monolingual group; there werethoSe individuals whd had good command of read-ing and writing the Spanish language as well asthose who had had no formal education in thatlanguage and were therefore unable tar understandthe Spanish directions in written form. Within, thebilingual group those individuals educated in theUnited Slates received no formal training in -theSpanish language and, although they spoke Span-ish, they were- unable to understand the writtencontents. Further, within this bilingual segment ofthe Spanisvh-speaking fiopulation, tfierc were thosewith a deficient education in, English. As a consequence, they cotild not be,adininistered the GATEin either of its written forms.

B. Spanish Translation of the Generld Aptitude TestBattery

Reports based on the various try(iut sessions in-

dicated that the translated instructions into Spanishfrom. English were satisfactory and easily under-stood. The instructions for the apparatus tests didcontain words that were unfamiliar to many in thesample. However, there was agreement that,through association by watching the administratormanipulate the items in question, they were ableto understand the significance of the instructions, .

espcicially since these words were repeated severaltimes during the practice exercises. The panelmembers concurred that the translation was feasi-

.ble for use throughout the southwestern UnitedStates. Those changes or modifications made tothe translation of the instructions were to ensureclarity and simplicity for the greatest number ofSPanish-ispeakiag individaals residing in the south-western United States.

The conclusiOas reached by the panel membe'rswas that the revised translation was cffective andaccurate, suitable for use in standardization andin operations.

135

BROADENING APPLICABILITY OF THE BEAG

After the tryout of the experimental Spanish GeneralAptitude Test Battery was completed in the Southwest,consideration was given to the possibility of doing addi-tional development to broaden the applicability of theJests to Spanish-speaking individuals throughout themainland United States, and possibly Puerto Rico. Twosteps, were taken in 'this direction. First, copies of thefinal version of the directions resulting from the South-west tryout were sent- to the New York:. Illittois andPuerto Rico agencies for their comments. Second, it

was decided to include Puerto Rico and other State

agencies in addition to the Southwest states in the study"Comparability of Spanish and English ,Ionguage Ver-sions of GATB Part I and 4," providing' a broad datacollection base for this phase of the test, developmentproject. As a result of taking these two steps, it was posesible to (I) incorporate additional changes in the direc-tions to give them broader applicability in testingSpanish-speaking applicants from a variety of back -'grounds and (2) develOp a more comprehensive national,data base for the conversion tables 'resulting from thecomparability study.

14 7 '

COMPARABILITY OF SPANISH AND ENGLISHLANGUAGE VERSIONS OF GATB PARTS 1 AND 4(STUDY,/ 1)

INTRODUCTIONFOrms A and.B of the experimental Spanish languageversion Of the GATB had item content identical to theirEnglish language GATB counterparts extept for Part(Name-Comparison), Part 4, (Vocabulary) and Part 6(Arithmetic Reasoning). Part I had Spanish rather thanEnglish language names; Part 4 had Spanish rather thanEnglish' language vocabulary, and Part..6 was an item=.for-item Spanish language translation of GATI3 Part 6.In the cases of Parts 1 and 4, it could not be assumedthat raw scores on the Spanish and English languageversions were comparable. Very ,likely the difficultylevels of the Spanish and English language versions ofPart 1 differ because the names to be compared. weredifferent. Although an attempt was made at item-for-item, word-for:word English to Spanish translation inPart 4;iit is likely that differences in word usage resultedin differences in difficulty levels, of the two versions ofPart 4. Thus, studies of comparability of the Spanishand English language versions of Parts 1 and 4 were

- required in order to establish 'appropriate aptitude conLversion tables for these tests,The problem of comparability did not exist for the othertats. In the case of Part 6, the change from the Englishlanguage version was translation of the. directions foradministration and the test items themselveS, but with-out change in essential meaning of the items. In thecase of the other tests, the only change was translationof the directions for administration. The items them-selves had no verbal content and remained the same asin their English language counterparts. Since is was rea-sonable to 'assume that no changes in difficulty levelresult from the English-to-Spanish translation, the sametest-to-aptitude conversion tables would apply as in theEnglish language GA-TB.

PROCEDUREEach participating State agency was asked to Obtainsamples of bilingual individuals who could pass bothSpanish and English language alternate forms of theGATB Screening Exercises, consisting of sample items

from Part 3TVee Dimensional Space, and Part- 4-Vocabulary of the GATB. Two samples were to be ob-tained by each participating State agency, as follows:

Sample A: At least SO high school 'senicirs. The ob-jectiye was to obtain ti,'samPle of individ-uals who would be able to take the GATBin either Spanish or English without dif-ficulty.

Sample 13: At least 30 individuals. age 17 or olderwith less than '12 years of education. Theobjective was to obtain. a less educatedbilingual group of individuals who have atleast a minimam amount of literacy in bothEnglish and Spanish. Possible sources ofindividuals included employed workers,MDTA trainees, local office applicants,WIN or CEP clients.

At each testing site the group to be tested was to bedivided into equal sized Groups 1 and 2. This divisionwas ft-) be done at random, but separately for males andfemides so that each group would have the same num-ber of males and the same number of females. Onepossible technique suggested was to make separate al-phabetically arranged lists of the names of the males andfemales and make alternate assignments to Groups 1and 2 starting at the top and working down for each list.Alternate approaches were permitted so long as therewas no danger of a systematic bias that would lead toarriving at groups that would not be comparable.' Thecritical importance that (1) a random procedure be es-uiblished in advance of any allocation to Group.1 or 2and (2) this procedure be followed rigidly in making theallocations was stressed.Once an individual had been assigned to Group 1 orGroup 2, he/she was to be retained in it for purposes ofdefinition of the group, data processing Sand analysis.Even if he/she subsequently dropped out before or dur-ing*e testing, the data available was to be tabulated.

hus, Groups I and 2 were to be equal in size, but thetwo groups would not necessarily have khe same number

9

of individuals with complete data.The testing schedule, identical for Samples A and B, wasas follows:

Order of Administration Group 1, Group 2First Special edition of Spanish

language Screening Exercises(Parts I and II) followed byalternate form (Form 'B) ofEnglish language ScreeningExercises (Part II only) fol-laWed by the Language lUsageIndicator.

. Second

Third

Fourth

Parts I and 4Form A(English)

Parts 1 and 4,Form B(English),

Parts I and 4BGPAForm A

The testing instruments were as follows: "A. GATB Screening Exercises: These Consist ofsam-

5le items from the; GATB vocabulary'and spatial

t sts. Their purpo'e is to deSermine whether an in-vidual has suffiCient literacy skills to take the

GATB.

Spanish language Screening Exercises,Part I: Translation of English language

Screening Exercises, PartI: SpatialPart II: Translation of English tanguage

Screening Exercises, Part II: Vocabulary.

English language Screening ExercisedPart II (Form B): An' alternate fOrtn of the

the English language Screening' Exer-`cises, Part II: Vocabulary

English language'GATB Tests i ,. Part 1, B-1002A and B: Alternate forms of

- Name Comparison Test, which measuresAptitude Q (Clerical Perception).

Part 4, B-1002A and B: Alternate forms ofVocabulary-Iest, which' measures AptitudeV (Verbal Aptitude) and G (GeneralLearning Ability).

Spanish language GATBPart 1, Experimental Spanish, Forms A and

B: Spanish name versions of GATB, B-1002A and B, Part 1.

Part 4, Experimental Spanish, Forms A andB: word-foi-word translation of GATB,B-1002A and B, Part 4.

Part 1, BGPA: Spanish name Puerto Ricanedition of GATB,.B-1002, Part 1.

Part 4, BGPA: Spanish word Puerto Ricanedition of GATB, B-1002, Part 4.

Language Usage IndicatorThis was an adaptation of a questionnaire used- ina study by Patella (1971). It was designed to Ob-

tain information on Spanish language backgroundand usagek

Parts 1 and 4Form A(ExperimentalSpanish) B.

Parts 1 and 4Form B(ExperimentalSpanish)

Parts I and 4BGPA'Form A

The tests were to be administered with the NCS an=swer sheet in ,accordance with the standard directionsprepared for them. The tests themselves (second, thirdand fourth items in the order of administration) wereto be administered in a single test session. The Screen-ing Exercises (first item) and the Language. Usage In-dicator were not required' to be administered as part ofthe testing session.The tests were to be scored in accordance with standardprocedures. Scores were also to be obtained on eachset of the Screening Exercises. Data were. td be tab-ulated on 80-column tabulation sheets in accordancewith instructions provided.Each participating State agency was to prepare a tech-nical report describing the conduct of the study, includ-ing information on selection and characteristics of

samples, procedures used for randomization into thetwo groups, number of individuals tested at each site,number of individuals screened out through application

yof the Screening Exercises, problems in test administra-tion, and any other information that would be pertinentin interpreting the results of the study..After completion of data collection, data tabulation, andpreparation of the report, the materials were to be for-warded to the National Office. The National Office wasto do the data processing on combined samples andanalyze the results.

10

C.

D.

ests

SAMPLES

Table 2 shows the number of cases in the final samplesfor the States participating in the study..Note that, in addition to the five southwest States withlarge Mexican-American populations, two eastern Statesand Puerto Rico were included to sample the PuertoRican Spanish-speaking population in the East. An at-tempt was made to obtain data also in Florida to includeindividuals of Cuban origin, but this was unsuccessful.Table 3 shows descriptive statistics for age, educationand sex for the combined samples from all locations.

16

Table 2

Number of Cases in Final Samples(First Comparability Study)-

'State

ArizonaColoradoNew JerseyNew YorkCaliforniaPuerto RicoTexasNew Mexico

Total

Table 3

N

69

70

83

102

134

60

83

91,

692

Descriptive Statistics for the Combined Samples

a

High School

Group 1(N=206)

Seniors - Low Education Adults

Group 2(N=196)

Group 3(N=143)

Group 4(N=147)

Age

Mean 18.00 17.98 27.47 27.50

S.D. .99 .87 8.68 9.40

EducationMean 11.51 11.57 9.14 9.38

S.D. .59 .54 2.04 2.13

Sex

Percent Male 42 44 48 48

The statistics are shown separately for the four analysisgroups (Groups 1 and 2high school seniors; Groups3 and 4low education adults). Note that there is goodcomparability between Groups I and 2 and betweenGroups 3 and 4 with respect to age, education and sex.The Language Usage indicator provided measures ofSpanish and English language use in various settings,variables useful in establishing comparability betweenGroups 1 and 2 and between Groups 3 and 4. Table 4shows the percentage distributions of the four groupson the six,. Language Usage Indicator items.

An inspection of the data in Table 4 shows the bilingualcharacter of the sample, with some important differencesin language preference between the high school seniorand low education adults. Although both groups tendto prefer using Spanish when conversing with parents,there is a tendency for the high school seniors to preferEnglish when conVersingwith friends in the neighbor-hood. This is no doubt partly a function of the greateruse of English in schools by the high school seniorgroup. The form of Spanish used was predominatelyMexican, Mexican-American or Puerto Rican, with only

11

12

Table 4

Percentage Distributions on Languagp Usage Indicator Items

1.

Language UsageIndicator Items

High School

Group 1(N = 206)

Seniors Low Education Adults

Group 2,(N= 196)

Group 3 Group 4(N= 143) (N =147)`

Language usedspeaking withparents:

English 21 23 6Spanish 49 52 68 , 76Both 30 26 27 19

2. Form of Spanishyou & parents'peak: N.,

Central American 1 2 3 5Cuban - 2 - / 5 6Mexican 18 21 10 16Mexican,Arri<rjklin 48 47 41 31Puerto Rican 26 23 33 33South American 2 3 3 3Other 2 - 2 5 7

3. Language used withfriends in neighborhood:

English 51 49 16 16Spanish 13 14 36 46Both 37 36 48 38

4, Language spoken inschools:

English 62 , 61 45 41.Spanish 3 5 31 34Both 34 34 24, 24

5. Number of radioprograms listened toin Spanish!'

None' 28 32 17 16Some j. 66 62 56 52,

More.than half 4 5 13 17All 2 2 14 15

6. Number of Magazinesand papiers readin Spanish:

.2eNone- 46 43 32Some 48 48 52 55More than half 6 8 6 12All 0 1 9 7

18

12% of the samplAising another form. With respect tocomparability in language usage/between Groups 1 and2 guid between Groups 3 and ,t; it is apparent that quitegood comparability Was. achieved.An Other basis for establishing comparability for theOopyrisoh groups consists of performance in the

Spanish and English language screening Exercises.Table 5 shows the means and standard deviati9ns ofscores (number right) on the screening exercises used inthis study. None of the differences between Groups 1and 2 or between Groups 3 and 4 is significant.

Table 5

Means and Standard Deviations of Scores on Spanishand English Language Screening Exercises

ZO,GA' Screening

Exercises

'2 High School SeniorsGroup 1 Grou '$2(N =206) (N=196)

.. M SD M SD

Low Education AdultsGroup 3 Grotip 4-,

(N =143) (N=147)M SD M SD

Spanish Language:Part 2.67 .65 2.67 .64 2.57 .66 2.51 .70Part II-Vocabulary 3.46 .81 3.51 .82 3.48 .84 3.61 .71

. English Language:

Part II-Vocabulary 3.23 .82 327 .87 2.68 .99 2.71 .89

RELIABILITY. ANALYSISThe results, in terms of alternate form reliability datafor Parts 1 and; 4, are shown in Tables 6 and 7. Each of GATB Parts 1 and 4 vs. their BEAG.counterparts:

these tables provider a comparison of statistics (means,standard deviations and reliability coefficients) for

Table 6

Means and Standard Deviations of Raw Scores on Forms A and B of Parts 1 and4, and Product Moment Correlations Botween Alternate Forms for High SchoolSenior Group 1 (English language GATB) and Group 2 (Spanish languageBEAG)

Part M

Group 1(N = 206)

GATB-English

SD

Group 2(N=196)

BEAG-Spanish

SD

Part 1, Form A 45.4 10.4 40.6 978

.82 .78

Part 1, Form B 49.8 11.4 46.5 11.5

Part 4, Form A 17.0 - 14.2 5.8

.80 .67iAr

Part 4, Form B 16.2 6.4 12.2 5.0

co"19'

13

Table 7

Means and Standard Deviations of Raw Scores on Forms A and B of Parts 1 and4, and Product Moment Correlations Between Alternate Forms for Low Educi-tion Adults Group 3 (English Ibnguage GATB) and Group 4 (Spanish languageBEAG)

Group 3=143)

GATE- English

Group 4(N =147)

BEAG-Spanish

Part M SD .1141 SD

Part 1, Form A 30.1 10.3 33.2 11.5

.8/ .81

Part 1, Form B 39.3 11.9 37.3 12.7

Part 4, Form A 10.9 5.9 13.1 5.8

.81 .82

Park 4, Form B 10.4 5.1 11.2 5.8

With one exception, the alternate form reliability co-efficients are within .02 correlation points of .80 forboth English and Spanish versions of Parts 1 and 4.The exception is the .67 reliability coefficient forBEAG, Part 4 in the high school senior group, which isconsiderably lower than the reliability of .82 found inthe low education adult group.Additional data bearing on reliability of BEAG Parts 1.

and 4 is provided by the correlations between these testsand BGPA Parts 1 and 4 for Groups 2 and 4. Thesecorrelations are shown in Table 8.

Table 8

Product-Moment Correlations BetweenBGPA Parts 1 and 4 and

Their BEAG Counterparts

Group 2(N = 196)

Group 4(N=147)Part

Part 1, Form A .78 , .83

Part 1, Form B .85 .86

-Part 4, Form A .67 .81

Part 4, Form B .83 .84

These correlations are quite similar to those between. Forms A and B of the BEAG shown for Groups 2 and

4 in Tables 6 and 7.

.14

I,

We may conclude that the reliabilities of S anishguage versions of Parts 1 and 4 compare quite favorably,with reliabilitics of GATB parts] and 4, except that.sonic improvement in reliability would be desirable fOPart 4, BEAG, on high school senior groups.

ITEM ANALYSIS, BEAG PART 4Item analysis was conducted on all 180 Part 4 Sparii.sh:language items in BEAG Forms A and B. and 1303?k,Form A. This was done for the following easons:

1. To obtain an objective check on the adequacy' ofindividual BEAG items, which are word-foi4vord .translations of Part 4 GATB items. Such i:che'ck

.was desirable because of the possibilitteS, thattrandations of the words from English fo.,SO4nishwould result in critical differences in.: iigahing,leading to changing the character,of an itiiii!'fromone with a clear-cut correct answer tooilewhichhis no correct answer or one which has triOrF:thanone correct answer. InforMation,on difficulty. level

and relationship between item and total test:per-formance would provide an objective basis ,forevaluating adequacy of each test item.

To determine the extent to which the items appear,in order of ascending difficulty. Although theGATB Part 4 items were ordered empirically onthe basis of difficulty level, it could not be assumedthat the Spanish translation of 'these items wouldretain the same order for a Spanish-speaking sam-ple.

2o

3. To make it possible to substitute BGPA items forBEAG items with poor discriminating power,based on appropriate item statistics.

Tables 9 and 10 show the-distribution of difficulty anddiscrimination levels of the items in BEAG 'Form A,BEAG Form B, and the BGPA. All three tests show awide range in item difficulty level and k:an discrimina-tion level.

Table 9°

Distribution of Item Difficulty Levels (pValues) `for the Three Spanish LanguageVersions of Part 4Vocabulary

(Groups 2 and 4N =343)

BGPAp ValueBEAGForm A

BEAGForm B

.00.09 2 3 0

.10.19 5 11 4

.20.29 13 11 13

_.30.39 6 12 10

.40.49 s 6 8 11

4 6 8

.60.t9 4 3 t7

.70.79 '4 4 N

.80 .89 4 2 1

.90.99 v 2 , o 1

Totals 60 60 60

Six of the items in BEAG Form A and three items inBEAG Form B have negative or .00 discrimination in-dexes. An attempt was made to replace the BEAG prob-lem items with 4,G.PA items which showed adequatediscrimination poWer end had similar qifficulty levels.Table 11 shows the statistical data supporting the rec-ommended substitutions.The data in Table 11 indicate that it was possible tosubglitute BGPA items with good discriminating powerfor BEAG items with .00 or negative discriminatingpower.It was not possible to achieve a one-to-one matching ofeliminated and substituted items with respect to diffi-culty level, but the net effect of the substitutions on themean and standard deviation of scores would not belarge. (Since-the difficulty levels of the substituted itemstend to be somewhat)wer than difficulty levels of theeliminated items, hi average scores would be slightlyhigher for th evised Part 4. The lower difficulty levels

Table 10

Distribution of Item Discrimination Levels(R,,,) for the Three Spanish LanguageVersions of Part 4Vocabulary

1(Groups 2 and 4N1=- 343)

Rp

trBEAGForm. A

BEAGForm B BGPA

.20.11 (neg) - 2 2 0

.10.01 (neg) 3 1

.00.09 5 2 2

.10.19 5 11 3

.20.29 10 11 5

.30.39 11 1,2 14

.40.49 12 13 20

.50.59 4 8 13'

.60.69 0' 0. 14' '-

Totals 60 60 66

together with the probable higher intercorrelation ofthe substituted items would tent] to decrease slightly thestandard deviation of the revised Part 4.)

DEVELOPMENT OF CONVERSION TABLESData from the high school senior Groups rand 2 wereused as the basis for development of tables for convert-ing raw scores on BEAG Parts I and 4 to equivalentscores tiit GATB Parts 1 and 4. Separate Group 1 andGroup 2 percentile distributions of scores on Forms Aand B of Part Land Part 4 were obtained and equiper-centile graphs were plotted for each of the four vari-ables (Part I-Form A, Part 4 Form A, Part I FromPart 4 Form B). Linear equations based on Group 1and Group 2 means and standard deviations of the fourvariables (see Table 6) were also developed and thecorresponding lines were superimposed on the equiper-centile graph.The results were nearly identical, indicating that alinear conversion of scores from Spanish to Englishequivalents would be appropriate.Table 12 shows the values for "a" and "b" used toconvert raw scores on BEAG Parts 1 and 4 to equiva-lent scores on GATB Parts 1 and 4.The aptitude scores corresponding to each BEAG Part 1and Part 4 raw score were obtained by (1) substitutingthe BEAG raw scores in the appropriate BEAG -GATBtest conyersion equations, (2) solving for the GATBequivalents, and (3) noting the GATB aptitude scoresderived from these raw score equivalents. The tables for

21

15

di,

..11Table 11

.

Proportion of the sample passing the Item (p) anti Point -Biserial Correlation ,(11 H) Between Item and Total Score for Part 4 Items Eliminted from theBEAG and Part 4 Items Substituted from the BGPA.

Eliminated BEAG Part 4 Items Substituted BGPA Part 4 ItemsForm Item No. P Rp bi. Item No. RP Ms

A 10 .16 -.10 48 .24 .52

A 26 .41 -.10 31 .23 .12

A 43 t' .26 .00 44 .32 .44

A 46 .12 .06 401 .25 .28

A 54 .11 8 .33 .40

55 .13 .00 57 .32

B 27 .02 -.14 g 56 .16 .33

48 .10 .04 13 .13 .18

B 55 -.10 15 .19' .46

converting raw scores on BEAG Parts 1 and 4 to apti-tude scores are shown in section I of the BEAG Mhn-ual.

Table 12

Values of "a" and "b" for use in EquationaX,.+ b for Converting Raw Scores on

BEAG Parts 1 and 4 (X,) to EquivalentScores on GATB Parts 1 and 4 (X,)

a bPart

Part 1, Form A 1.059 2.377

Part 1, Form B .986 3.958

Part 4, Form A 1.063 1.850Part 4, Form B 1.274 .619

Mean Aptitude \' (Verbal- Aptitude) and Q (ClericaiPerception) scores for the high school and low educa-tion adult samples are shown in Table 13. These meanscores indicate that (1) comparability of aptitude scoresfor the high school senior Groups 1, and 2 has beenachieved and (2) low education adults tend to scoresomewhat better on the BEAG than on the GATB.

DISCUSSIONAs indicated in Table -13, the bilingual high schoolseriior groups have identical average scores on Spanish

16

-and English language editions 'of the GATB forAptitude V and 'for Aptitude Q. Howevg, it is notnecessarily true that a given individuals Ares wouldhe the same or nearly the same on the Spanish andEnglish editions. The correlations between f3GPA andGATB measures of Aptitude 0 (.76 and .81) are highenough to accept the scorts as adequate measures ofAptitude Q whether administered in Spanish or inEnglish. But this is not the.casefor Aptitude V becausecorrelations between BGPA and GATB measures ofAptitude V (.22 and .09) are much too low to haveconfidence that results of testing in the Spanish andEnglish editions would not differ substantially for bi-lingual high school seniors. Thus, Spanish-speaking highschool seniors who have been educated primarily in theEnglish language should ordinarily be tested wiartheGATB rather than the BEAG. Howevdr, when individ-uals who have had substantial English language educa-tion are tested with the BEAG, Forhi A, it may beuseful to administer, in addition, the GATB, Part 4,Form B, C,, or D as a check oil Verbal1Aptitude. Bi-linguals with low education should ordinarily be testedwith the BEAG because their Aptitude V and 0 scoresaverage higher when tested with the BEAG than when

tested with. the GATB.Based on the research conducted to develop it, includ-ing the attempt-to broaden its applicability to individualswith a variety of Spanish language backgrounds, theBEAG appeared ready for nationwide operational use.However, the following additional research was needed

ti

rove the BEAG and to evaluate its adequacy inlocal office operations:

1. The low education bilingual adults to d to scorehigher on the BEAG thjin on the GATE for both'Parts 1 and 4. This was to be expected, and pro-video a basis for using the BEAG for bilingualswhen they have less than a high school education'However, the GATB-BEAG differences in averageAptitude Q scores are 9 points for Form A Butonly 3 points for Form B. An additional checkstudy on Part I would be useful to check on theconversions.

a

fr2. The substitution of strong BGPA items for weak

BEAG items improv,es the BEAG's Part 4 meas-urement .qualities but may have some measurableeffect on the BEAG -GATB Part 4 conversions. Inaddition, the BEAG items are not in strict orderof increasing difficulty. The present study shouldbe replicated using the revised BEAG Part 4,Forms A and B, with items listed in order of dif-ficulty.

3. nationwide owational evaluation. (-)f the BEAGshould be conducted to determine whether addi-tional chalittes in the directions for administrationand use of the tests are needed.

Table 13

Mean Aptitude V and Q Scores for High chool Seniorand Low Education Adult Samp s

Part Form Aptitude

High School SeniorsGroup 1 Group 2

(N = 206) (N = 196)

Low Education AdultsGroup 3 .(N =143)

Group 4(N =147 r*

1 A 0.--\

107' 107 89 98

1 B '136 116 117 120

4 A V 94 94 82 92

4 B V 92 92 81 90

ti 17

I

COMPARABILITY OF SPANISH AND ENGLISHLANGUAGE VERSIONS OF GATB PARTS 1 AND 4(STUDY 2)

INTRODUCTIONResults 'from Comparability StkIdy I provided a basisfor tables to convert raw scores on .Parts I and 4 of theBEAG to aptitude scores. These conversion tables weyeused in a nationwide operational: tryout of the BFAGconducted to determine what revisions would he de-sirable in the directions fOr administration. it was alsodecid. to conduct a second study on eomparability ofBEAG and GATB Parts I and 4. The reasons for asecond study were as follows:

1. The BAG 'Part 4--Vocabulary measure used inthe oritinal study was item analyzed, and strong

items from the Puerto Rican Bateria General dePruebas de Aptitud (BGPA) were 'Rubstit4ted forweak BEAG items. The measurement characteris-tics of the revised test, with items reordered bydifficulty level, could have some Measurable effecton (1) conversions of BEAG Part 4 scores toAptitudescores and (2) reliability of measurementas indicated' by the correlation betwicen scores onForm A and Form B.

2. The average GATB-BEAG score difft!rence onAptitude Q for a low education 'bilingual groupwas 9 points for Form A but only 3 points forForm B. A second study was desirable to makeanyzidjustments that may be necessary in,the con-versions.

'PROCEDUREThe procedure was much the same as that used for the'high school senior portion of the original study.As in the original study, each partieipating'S ate agencywas to obtain a sample of at least-50 Hint ual highschool seniors. The objective was to obtain a si mple ofindividuals who would he able to take the TB irr-either Spanish or English without difficulty.The procedure for dividing the sampl o be tested ateach site into equal sized Groups nd 2 was the same_

as in the original study. ThiS- division was to he chineat random, but separately for males aticbtemalcs sK7 got.each group would have the same number <if males andthe same number of females.

The teseing,-schedulc was as follows:

'Order ofAdministration Group 1,

First' ' Parts I. and 4

N

Second

Form U (GATT)Parts I and 4Form A (GATB)

Group 2

Parts 1 and 4Form B (BEAD)Parts 1 and4,4"Form A (BEAG)

Note that the order of administration (Form 11----FormAl is the reverse of the order in the original study.

tine instruments used were the same as in theI study except that (I) the GATB Screening Ex-.

and the I .atigu age Usage Indicator were notused. (2) the item -anal zed versions of BEAG Part 4,Forms A and 13 were used, and (3) either the NCS orIntran answer sheet could he used.

The t

cercisc

SAMPLESTable 14 shows the number of cases in the final sam-ples for the States participating in the study.

Table 14

Number of ases in Final Samples(Second C mparsability Study)

,,State N

Arizotia

r,CaliforniaColorado

FloridaNew Jersey

New MexicoX.) Yorkexas

To al

33

84

SR

6Rr

51

60

,53

487

Note that, as in the first comparability study,possible Ski obtain representation from both Fast

it' wasand

lh

West. One of these eastern States was Florida, not rep-resented in the original study. All of the individualsfrbm Florida were.of Cuban origin,.Table 15 shows descriptive statistics for age, education,Iindsex for the combined samples from all locations.The two groups are comparable with respect to ageand sex charateristics.

Table 15

Descriptive Statistics for the CombinedSamples

Ate:

Group 1(N =250)

Croup 2(N=237)

Mean 17.0 I 16.96

S. D. 1.014\,

1.00

Sex

PerL'cnt Male \140 42

RELIABILITY ANALYSISThe results, in terms of alternate form reliability (1, tofor Parts 1 and 4, are shown in 'Fable 16. This table

rt/vides a comparison of statistics (means, standard

20

'deviations and reliability coefficients) for GATB PartsI and 4 vs. their BEAG counterparts.The reliability'. coefficients for the Spanish, languageBRAG Parts I and, 4 are in the .80's'and are the samesize as the reliability coefficients for GATB,Parts 1 and4,. We may conclude that the reliabilities of the finaleditions of BEM; rats 1 and 4 are excellent. .1

MODIFICATION OF CONVERSION TABLEThe results from the second- comparability study, interms of differences between means and standard devia-tions-iTh-scores on English atd Spanish Versions of Parts

I and 4,- arc quite similar to those from the first study,on which the aptitude conversion tables were bitsed. Theone exception relates to BEAG Part 41, Form B. Thestandard deviation, of scores. for this test was Mite low('.(1) in the original study, an in additiyn the rmA--Form B reliability of scores was considerably liowerfor BEAG Part 4 than for GA,TB Part 4. In the second

(study the standard deviations of BEAG and GATB,versions of Part 4, Form B fire more nearly the sameand the reliabilityof scores is the same (.86) for BEAGand GATB version of Part 4, Form B. Therefore, itappears reasonable to modify the conversion table forBEAG Part 4, Form B using data for thAecond study.This was done, and the revised equation for convertingraw scores on BEAG Part 4, Form B to equivalentSc'ores on GATB, Part 4, Form B is as follows:

Xe .931Xsrf 4.329)1.

Table 16

Means and Standard Deviations of Raw Scores on FOrms A and B of Parts 1 and4, and Product-Moment Correlations Between Alternate Forms for Group 1(Ehglish language GATB) and Group 2 (Spanish language BEAG)

Part NI

Part _1, Form B 41.5

'Part 1, Form A 50.2

Part 4, Form B 16:1

Part 4, Form A 17.1

Group 1(N =250)

CATII-English

SD

Group 2(N = 237)

/WAG-Spanish

SD

10.0

.R1

39.5 1'0.2

go

12.0 45.4 11,7

6.1 12.7 6.6k 86

6.3 14,4 6.0

rA., 0

COMPARABILITY. OF BEAG AND BGPA VERSIONSOF GATB PARTS 1 AND 4

INTRODUCTIONThe Baterfa General de Pfuebas de Aptitud BGPA )is the Puerto, Rican Spanish edition of the GATB whichwas used in Puerto Rico and also on the mainland tosome extent since' In 1965 a Puerto Rico stand-'ardization of the BGPA was 'completed. which resultedin test-aptitude conversion tables based on data from a,sample 'of the Puerto Rico General Working-Popula-tion.' This standardization results in .aptitude scoreswhich have a mean of 100 and a standard deviation of20 for the 'Puerto Rico General Working Populationsample (U.S. Department of Libor, 1965).Can the conversion tables developed for the 13GPA heused with BEAG tests? The answer to this questiondepends on the comparability of the tests i i the twoeditions.- It is reasonable to conclude that such corn-parability exists for all tests exceRtiPart 1 and Partwhich consist of different itervis in the two Sp,versions of the GATB.In the course of conduc-ting the first BEAG-GATB coin-parability,study on Parts 1 and 4, data were collectedy Descriptive Statistics for Puerto Rico Samplepertnitting comparison of BEAG and BGPA score Ofindividuals in the samples obtained. These data °videsome evidence that raw scores on BEAG Parts 1 and 4are comparable to scores on BGPA Parts 1 ,and 4,respectively. however, the Puerto Rico sample for thestudy was not large and there may be some question ofwhether effects of practice were controlled sufficientlyto permit firm conclusions in this regard:. Accordingly,a special study was designed to provide the basis for ananswer to this question:.

Group 1 . Group 2

BEAG Parts 2-3 BEAG--Parts 2-3BEAG Part 4 BGPAPart, 4BGPAPart 4 BEAGPart 4BEAGParts 5-12 BiAG-7Parts 5-12

Random procedures were used at each office for alloca-tion of counselees to Groups 1 and 2, and this was doneseparately for males and females. The tests were ad-ministered by a professional counselor at each of thetwo testing sites. The number tested was 50 in Group

and 50 in Group 2, with, each group consisting ofequal numbers of males and females. Table 17 showsthe sample characteristics for the two... groups. Thesedata indicate that the two samples are comparablewith respect to age, education, and sex characteristics,

Table 17

PROCEDUREThe study was conducted in two locations. Counseleeswere tested with the entire BEAG for purposes of op-/erational tryout and With the BGPA Parts I and 4 forthe. purpose 'of ob aining data on BGPA and BEAUcomparability for these two tests. To control the effectsof practice, the eounselees were divided randomly intotwo groups and tested in the following order:

Group 1

BEAGPart IBGPAPart I

Grsoup 2

BGPAPart 1BEAGPart 1

1.4

Group 1(N=50)

Group 2(N=50)

Age

Mean,S. D.

22.606.29

22.605.51

Education

Mean 11.58

S. D.'. 1:17 1.04

Sex

Percent .Male 50 50

RESULTSTilbk 18 shows the-result in terms of a comparison ofGroups l and 2 with respect to means, standard devia-tions and reliability coefficients of Parts 1.. and 4.

A' comparison of the 'statistics for the two groups'in-

21

dicates that there is good comparability between BEAGand BGPA versions of Parts 1 and 4. Although it ap-pears from these data that BEAG Part 4 may be slightly-.easier than BGPA Part 4, no such difference was found

in the first BEAG-GATB comparability study. There-fore, do adjustment appears to be required in either

..Part 1 or --Part .4 ,conversion.,tables.for..,PuertoRicorwhen the BEAG is used there.

Table 18

Means and Standard Deviations of Raw Scores on BGPA and BEAG Parts 1 and4, Fornil, and Product Moment Correlations Between BGPA and BEAG Edi-tions of Mese Tests for Groups 1 and 2

Group(N=50)

Part M SD

BEAG Part 1 32.8 10.8

.75

BGPA Part 1 10.5

BEAG Part 4 14.4 4.9

87

BGPA Part 4 15.9 8.1

eart

Group 2(N = 50)

M SD

BGPA Part 1 32.0 11.6

.74

BEAG Part 1 41.1 9.9

,BGPA Part 4 12.5 6.1

.82

BEAG Part 4 15.2 6.0

NATIONWIDE OPERATIONAL TRYOUT OF THE BEAG1

TRYOUT PLAN SpanishIspeaking applicants or because they lacked

The original design developed by the N;itional Office qualified' Spanish-speaking test administrators.

called for the evaluation of the BEAG to be conductedRESULTS AND CONCLUSIONSconcurrently with its, operational use in State ES offices

throughout the country. Every State, agency which had Twenty State agencies administered the BEAG to more

been using the BGPA or was serving Spanish-speaking than 2578 individuals during the evaluation period, as

applicants who would benefit from testing with the shown in Table 19. The other States with BEAG mate-

BEAGiwas asked to participate. Testing sessions were rials did not install the BEAG for various reasonsto be conducted by experienced, Spanish-speaking test not, nough Spanish-speaking applicants who could

administrators. Beginning in September 1974, test benefit from BEAG testing, staffing problems, agency

administrators were instructed to: (1) administer and budget cut-backs, changes in agency priority services,

score the BEAG in accordance with the directions and etc. Nonetheless, the tested group represented a cross

procedures in the May 1974 Spanish language edition section of the Spanish-speaking community through-

of the Manual for the GATB, B-1002, Section I, (2) out the mainland and Puerto Rico.

note testing problems encountered in each testing ses- The detailed reports submitted by the 20 participatingsion. (3) keep a record of the number of individuals State agencies ,supported the conclusion that it wastested with the BEAG, and (4) recommend improve- desirable and feasible to develop a revised set of ad-ments in the directions for administering the test bat- ministration directions and procedures to meet. thetery. needs of BEAG users on the mainland and in Puerto-At the end of the evaluation 'period in July 1975, each Rico. A test research team from the New York Stateparticipating State agency was to prepare a report con- Employment Service worked on the project of synthe=solidating this information and forward it Rohe Na- sizing the reports in the early months of 1976. Theirtional Office. These reports would, in turn, be reviewed recommended/ revised edition, and a report indicatingto develop a,final set of administration directions' and the basis for making each specific change, were sentprocedure. which would meet the needs of all BEAG to each State agency which had Made substantial com-users. ments on the tryout, requesting final comments. These

comments were then incorporated in an August 1976DEVELOPMENT OF MATERIALS draft which was given a final intensive review by ex-The Test Research Center of the California State Em- perts in the California, New York, and Puerto Ricoployment Service prepared the BEAG materials for agencies. A final consensus was achieved at a Decem-operational use. These included the May 1974 Spanish ber 1976 conference in Washington.language dOion of the Manual for the GATB, B The final BEAG directions for administration have an1002, Section I: Administratirth and Scoring: Test extensive research base, derived from careful initialBooklets I & II. Forms A & B; and Part 8. State

agencies had the option of choosing the Spanishconstruction, tryout and review; revision based on na-

language answer sheet developed by NCS or INTRANtionwide tryout on a variety of Spanish-speaking

Corp. for use with the BEAG. groups; and final concurrence of Spanish-speakingteams of technical experts from California, New York.

PARTICIPATING STATES and Puerto Rico. As such, we may conclude that these

Thirty-one State agencies and the District of Columbia directions will be suitable throughout the mainlandManpower Administrition requested BEAG smaterials. and inePuerto Rico as the standard set of directionsThe remaining State agencies did not participate either for administering the new standard Spanish language

because they did not have a sufficient. number of version of the GATB.

2s23

24

Table 19

BEAGs Administered during Nationwide Tryout:September 1974,July 1975

Regi 9n State BEAGs SATBs

II PUERTO RICO 1249 264NEW YORK 53NEW JERSEY 303

III, ,DISTRICT OF COLUMBIA 291

IV FLORIDA 20

V ILLINOIS 36MICHIGAN 23WISCONSIN 9MINNESOTA 2

VI NEW MEXICO , 45TEXAS 100 68LOUISIANA 6

. VIII COLORADO* +WYOMING 3 '\

IX NEVADA , \ 5444.. ARIZONA 25

CALIFORNIA* +

X WASHINGTON * N +IDAHO * +OREGON 26

Totals 2245 + 332

* Tested, number not reported.

REFERENCES

Patella, Victoria M. "A study of the validity of lang-uage usage as an indicator of ethnic identification."M. A. Thesis, Texas A&M University, 1971.U.S. Department of Labor. Standardization of the Gen-eral Aptitude Test Battery for Puerto Rico. USES Spe-cial Technical Report No. 8. U.S. EmploymentService, Washington, D.C., 1965.

if U.S. GOVERNMENT PRINTING OFFICE: 1978 0-245-438

30

t

25,

A